Structure learning of Bayesian networks using a semantic genetic algorithm-based approach